Brains Trust

What Is Brains Trust?

Brains Trust is a Claude Code skill designed to streamline collaborative decision-making by leveraging the expertise of multiple leading AI models. Rather than relying on a single viewpoint, Brains Trust enables developers and technical teams to consult with various large language models (LLMs) from providers such as OpenAI (GPT), Google (Gemini), and others through the OpenRouter API. The skill can be invoked for code reviews, architecture decisions, strategy discussions, prompt engineering, debugging, and more—any scenario where a "second opinion" or even a consensus among AI experts might improve outcomes. Brains Trust is purpose-built for developers using the Claude Code environment, offering easy triggers and automation to blend AI perspectives into your workflow.

Why Use Brains Trust?

Modern software development often benefits from collaborative input, especially when tackling complex problems or making architectural choices. While AI assistants like Claude provide robust help, there are times when diversity of thought or a challenge to prevailing assumptions can surface better solutions. Brains Trust addresses this by:

• Reducing blind spots: Different models have distinct training data and biases, making their combined feedback more comprehensive.
• Speeding up peer review: When human reviewers are unavailable, AI consensus helps catch issues or highlight alternatives.
• Enhancing creativity: Devil’s advocate and multi-model patterns can provoke new lines of thought.
• Validating strategies: Quickly compare approaches, from coding patterns to infrastructure decisions, across AI perspectives.
• Minimizing single-model limitations: No single LLM is infallible; cross-referencing models increases confidence in recommendations.

How to Get Started

Brains Trust is designed for ease of use within the Claude Code environment. To start using it:

1. Install the Skill:
   Clone or copy the skill from the Brains Trust GitHub repository into your Claude Code environment’s skills or plugins directory.

2. Configure API Access:
   Ensure you have API keys or access credentials for the LLMs you wish to consult (e.g., OpenAI GPT, Google Gemini, or via OpenRouter).

3. Invoke the Skill:
   Trigger Brains Trust via any of its supported phrases, such as:

   • brains trust
   • second opinion
   • ask gemini
   • ask gpt
   • peer review
   • consult
   • challenge this
   • devil's advocate
   • what does gemini think
   • what does gpt think

   You can specify a question or topic, or simply invoke with the keyword to consult on the current context.

Example Usage:

brains trust Should we use MongoDB or PostgreSQL for this microservice?

Or, to review code:

ask gpt Review the following function for security issues:
def process_user_input(input_string):
    eval(input_string)

Key Features

• Multi-Model Consensus:
  By default, Brains Trust consults at least two different LLMs and presents a consensus or contrasting viewpoints. This simulates a real-world "brains trust" panel.

• Context Awareness:
  If invoked without a specific topic, Brains Trust analyzes recent session activity—such as recent code edits or architectural discussions—to determine what to review or discuss.

• Pattern Selection:
  Supports several consultation patterns:

  • Consensus: Collate answers from multiple models.
  • Devil’s Advocate: Request one model to challenge the prevailing solution.
  • Single Opinion: Ask a specific model (e.g., Gemini or GPT) for its view.

• Flexible Triggers:
  Multiple trigger phrases make invoking the skill natural and adaptable to different workflows.

• Broad Scope:
  Not limited to code. Use for architecture, design, strategy, prompting, debugging, and writing.

• API Agnostic:
  Works with any models accessible via OpenRouter, Gemini, or OpenAI APIs.

Best Practices

• Be Specific When Needed:
  For critical reviews, specify exactly what you want feedback on. For example, “brains trust Review the error handling in the payment module.”

• Use Contextual Invocations:
  When working through a complex problem, trigger Brains Trust after major decisions or before finalizing changes.

• Compare and Synthesize:
  Review outputs from all consulted models, noting points of agreement and divergence. Use these insights to inform your final decision.

• Automate in Workflows:
  Integrate Brains Trust into CI/CD pipelines or code review processes for automated AI-driven peer review.

• Challenge Assumptions:
  Utilize the “devil’s advocate” pattern to intentionally surface alternative perspectives, especially for strategic or architectural decisions.

Important Notes

• Privacy and Security:
  Information sent to third-party APIs (OpenAI, Gemini, etc.) is subject to their privacy policies. Avoid sharing sensitive or proprietary code unless compliant with those terms.

• API Costs and Limits:
  Each model queried may incur usage charges or rate limits. Configure your API keys and quotas accordingly.

• Model Biases:
  Different LLMs may have differing strengths, weaknesses, or domain expertise. Use Brains Trust as an augmentation, not a replacement, for expert human judgment.

• Claude Code Compatibility:
  Brains Trust is currently compatible only with Claude Code environments.

Brains Trust empowers teams and individuals to elevate technical decisions by tapping into the collective intelligence of today’s leading AI models. By integrating it into your workflow, you gain access to richer, more reliable insights—enabling higher quality code, robust architectures, and better outcomes.